LOGICPRPBANK

LOGICPRPBANK是由匹兹堡大学和埃默里大学联合创建的一个包含7093条命题逻辑语句的数据集，涵盖了代数、算术、微积分、几何、数论和统计学六个数学领域。该数据集旨在研究语言模型在逻辑蕴含和等价推理方面的能力。创建过程中，使用了ChatGPT生成原子命题逻辑语句，并通过人工验证确保其准确性。LOGICPRPBANK的应用领域主要集中在智能教学系统中，用以提升机器在数学逻辑推理方面的表现，解决传统语言模型在此领域的不足。LOGICPRPBANK is a dataset comprising 7,093 propositional logic statements, jointly created by the University of Pittsburgh and Emory University, covering six core mathematical disciplines: algebra, arithmetic, calculus, geometry, number theory, and statistics. This dataset is designed to investigate the logical implication and equivalence reasoning capabilities of language models. During its development, atomic propositional logic statements were generated using ChatGPT, and their accuracy was ensured through manual validation. The primary application scenarios of LOGICPRPBANK are focused on intelligent tutoring systems, where it is used to improve machines' performance in mathematical logical reasoning and address the shortcomings of traditional language models in this field.

搜集汇总

背景与挑战

背景概述

LOGICPRPBANK是一个由匹兹堡大学和埃默里大学创建的命题逻辑数据集，包含7093条语句，覆盖六个数学领域，旨在评估和提升语言模型的逻辑推理能力，主要用于智能教学系统。